Introduction

About Me

  • BSc Economics (UCL)
  • MA Publishing (UCL)
  • MSc Business Analytics (Imperial College London)

Structure

  • Context
  • Methodology
  • Results
  • Application
  • Conclusion

Context

Context

The current forecast is based on the following criteria:

  • Past sales of an author (or comparable authors, for a debut title)
  • Recent performance of a book's genre/category for a publisher
  • Potential for foreign and subsidiary rights sales
  • Whether the title was auctioned

Context

Non-Fiction Miss
Income
Gross Sales 1204352
Returns -256800
Net Revenue 947552
Book Cost
Cost of Sales -188266
Royalties -301809
Gross Margin 457477
Provisions
Advances -640416
Stocks -21410
-661826
Subrights 7825
Gross Profit -858350
Selling Cost
Marketing & Publicity -46561
Customer Support -34137
Distribution -56853
Comimission -9476
Contribution -1005377
Overheads -50000
Loss -1055377
* Slide Produced by a Top-5 UK Trade Publisher

Context

Simple Random Walk Model
A model which presumes that the current value is determined by the past value and a random error term.


More Formally
\(y_{t} = y_{t-1} + e_{t}\)



The Main Assumption
e(t) has to be zero on average.


No specific reason for this to be true for sales forecast.

Context

Research Objectives

  1. What Are the Factors That Determine the Sales of a Book?
  2. What is the Quantitative Impact of These Factors?

Methodology

Methodology

Strategy

Price and Quantity were analysed separately.

  1. Price
    • Primary Research
      • 500 self administered surveys
      • Stratified sampling approach
      • Hypothesis Testing
  2. Quantity
    • Secondary Research
      • Sample size of 34000 (obtained from Nielsen BookScan)
      • Regression Analysis

Methodology

Primary Research - Geographical Scope

Methodology

Why use Regression Analysis?

Suppose we have sales data of 1000 books and we can separate these books in two groups with the following characteristics:

Factors Group 1 Group 2
Sales Quantity of Each Book 100 50
Metadata Complete Incomplete
Author's First Name Does not start with an 'A' Only starts with an 'A'
Colours on Cover Page 2 More than 2

Results

Results

Price

Stronger quality signalling factors have a stronger impact on consumer's willingness to pay.

Factors Impact on Consumer's Willingness to Pay (%)
Winning an Award 23.5
Nomination for an Award 17.6
5 Star Amazon Rating 15.7
Bestseller Status 14.9
Movie Adaptation 12.7
Publisher's Brand 4.3

Results

Price

Pricing strategies can be set according to customer responsiveness. Tesco customers are much more responsive to the quality signalling factors in comparison to Waterstones.

Results

Quantity

Dependent variable:
Log(Sales Volume)
Pre-Release Post-Release Full
Book Author Signalling Final
I(log(asp)) -1.445*** -1.123** -1.010** -1.086**
months 0.060*** 0.040***
I(months2) -0.0002*** -0.0002***
genreChildren 4.262*** 4.617*** 3.961*** 3.792***
genreFiction 4.973*** 6.774*** 5.247*** 4.998***
genreNon-Fiction 4.865*** 5.510*** 5.624*** 5.474***
hardback1 -1.058** -0.745 -0.980** -1.062***
famous_pub1 1.666***
colors -0.040
I(log(size)) 0.836*
I(log(facebook)) 0.220***
I(log(twitter)) -0.001 0.104**
native1 0.607
book_exp 0.001
award1 1.952*** 1.587**
movie1 0.929*
bestseller1 3.463*** 2.930***
Adjusted R2 0.642 0.554 0.702 0.727
Note: [*] p<0.1; [**] p<0.05; [***] p<0.01

Results

Quantity

  1. Post-Release Signalling Factors
    • Bestseller Status has the highest positive impact of 293% on sales, followed by the impact of winning an award (158%).
  2. Author Specific Variables
    • Twitter following has a significant impact on book sales. On the other hand, the impact of Facebook likes was not significant.
    • Interestingly, the number of books previously written by an author and the author's ethnicity does not have a statistically significant impact.

Applications

Applications

Direct Implications

Pricing Strategies

  1. Prices can be increased when a book is associated with any of the quality signalling factors* mentioned in this study.
    • The increase in the price can be proportional to the increase in consumer's willingness to pay.
  2. Discounts given to the retailers can be optimised based on the responsiveness of the customers to various factors like bestseller status, movie adaptation, winning an award, etc.
  3. The return from quality signalling factors can be aligned with the investment and risk associated with the factor.



*Quality signalling factors include bestseller status, winning an award, nomination for an award, movie adaptation, Amazon ratings and publisher's brand.

Applications

Direct Implications

Demand Forecasting and Efficient Resource Allocation

  1. Regression can help control for various factors and can provide a far accurate prediction compared to just using the past sales.
    • Can help avoid giving huge advances which can be difficult to recover.
    • Increase competitiveness by allowing the publisher to give as high advances as financially feasible.
  2. Useful for analysing the most effective marketing channels
    • Can be used for allocating budget between various marketing channels based on their relative effectiveness.
    • For instance, if online media appears to be more effective than offline media and if Twitter has the highest impact on book sales, then marketing campaigns can be focused towards Twitter.
  3. Can help identify the importance of various awards and their impact on sales

Applications

Possibilities

  1. Analysing manuscripts
    • There are machine learning algorithms which can easily help to identify topics, complexity and many other features.
  2. Building Classifiers
    • This information along with the metadata information can be used to create an algorithm which can help the publishers in their decision of publishing a book.
  3. Analysing Macro-Level Sentiments
    • By using tweets it is possible to understand if there is a positive or negative sentiment about a particular topic.
    • This information can be used to find opportunities and gaps in the market.

Conclusion

Conclusion

It is time for the publishing industry to make use of the available data to make smart business decisions and move beyond the gut-based decision-making process.

In God we Trust, all others bring data - W.Edwards Deming

Thank You

Contact Information